Characterization of mice lacking candidate N-acyl ethanolamine

biosynthetic enzymes provides evidence for multiple pathways that

contribute to endocannabinoid production in vivowzGabriel M. Simon and Benjamin F. Cravatt*

Received 7th January 2010, Accepted 19th March 2010

First published as an Advance Article on the web 14th April 2010

DOI: 10.1039/c000237b

The biosynthesis of the endocannabinoid anandamide (AEA) and related N-acyl ethanolamine

(NAE) lipids is complex and appears to involve multiple pathways, including: (1) direct release

of NAEs from N-acyl phosphatidyl ethanolamine (NAPE) precursors by the phosphodiesterase

NAPE-PLD, and (2) double O-deacylation of NAPEs followed by phosphodiester bond

hydrolysis of the resulting glycero-phospho (GP)-NAEs. We recently identified GDE1 as a

GP-NAE phosphodiesterase that may be involved in the second pathway. Here, we report the

generation and characterization of GDE1(�/�) mice, which are viable and overtly normal in their

cage behavior. Brain homogenates from GDE1(�/�) mice exhibit a near-complete loss of

detectable GP-NAE phosphodiesterase activity; however, bulk brain levels of AEA and other

NAEs were unaltered in these animals. To address the possibility of compensatory pathways,

we generated GDE1(�/�)/NAPE-PLD(�/�) mice. Conversion of NAPE to NAE was virtually

undetectable in brain homogenates from these animals as measured under standard assay

conditions, but again, bulk changes in brain NAEs were not observed. Interestingly, significant

reductions in the accumulation of brain NAEs, including anandamide, were detected in

GDE1(�/�)/NAPE-PLD(�/�) mice treated with a fatty acid amide hydrolase (FAAH)

inhibitor that blocks NAE degradation. Finally, we determined that primary neurons from

GDE1(�/�)/NAPE-PLD(�/�) mice can convert NAPEs to NAEs by a pathway that is not

preserved following cell homogenization. In summary, combined inactivation of GDE1 and

NAPE-PLD results in partial disruption of NAE biosynthesis, while also pointing to the existence

of an additional enzymatic pathway(s) that converts NAPEs to NAEs. Characterization of this

pathway should provide clarity on the multifaceted nature of NAE biosynthesis.

Introduction

The endocannabinoid system plays a key role in both central

and peripheral tissues, where it participates in diverse physio-

logical processes including nociception, inflammation, appe-

tite, and metabolism.1–4 This system is comprised primarily of

the cannabinoid G-protein coupled receptors CB1 and

CB2, their two endogenous arachidonate-containing lipids

(‘‘endocannabinoids’’) anandamide (N-arachidonoyl ethanol-

amine, AEA) and 2-arachidonoyl glycerol (2-AG), and the

enzymatic pathways responsible for endocannabinoid bio-

synthesis and degradation.5,6

Classical neurotransmitters, such as amino acids, mono-

amines, and neuropeptides, are water-soluble and are stored in

vesicles prior to release into the synaptic cleft upon neuronal

depolarization. In contrast, lipid messengers such as the

endocannabinoids are considered to be too hydrophobic for

storage in vesicles and are therefore believed to be synthesized

‘‘on demand’’ from phospholipid precursors.7 The signaling

potential of endocannabinoids is consequently under strict

regulation by the enzymes and proteins that catalyze their

biosynthesis, transport, and degradation.5,8–10 The identifica-

tion and characterization of these enzymes is crucial to obtaining

a complete understanding of endocannabinoid signaling and

its role in mammalian physiology, as well as to ascertain the

potentially distinct functions played by AEA and 2-AG.

AEA is a member of a large class of amidated lipids termed

the N-acyl ethanolamines (NAEs)11 and is the principal

member of this family with CB1 agonist activity.12,13 Early

studies on changes to the lipidome resulting from ischemic

shock in dog heart and brain tissue revealed rapid simul-

taneous accumulation of NAEs and an unusual phospholipid

species,N-acyl phosphatidyl ethanolamines (NAPEs), suggesting

a precursor–product relationship between these two classes of

lipids.14,15 Several alternative hypotheses concerning the bio-

synthesis of NAEs have been proposed including direct con-

jugation of ethanolamine with fatty acids or acyl CoAs;16,17

however, multiple lines of evidence point to NAPEs as the

endogenous precursors to NAEs.18 In the early 1980s, several

The Department of Chemical Physiology and The Skaggs Institute forChemical Biology, The Scripps Research Institute, 10550 N. TorreyPines Road, La Jolla, California 92037, USA.E-mail: cravatt@scripps.eduw This article is part of a Molecular BioSystems themed issue onChemical Genomics.z Electronic supplementary information (ESI) available: Three supple-mental figures are provided that describe additional in vitro and in situexperiments investigating NAE levels and NAPE metabolism. SeeDOI: 10.1039/c000237b

This journal is �c The Royal Society of Chemistry 2010 Mol. BioSyst., 2010, 6, 1411–1418 | 1411

PAPER www.rsc.org/molecularbiosystems | Molecular BioSystems

pathways were proposed by which NAEs could be released

from NAPEs, including direct hydrolysis by a D-type phospho-

lipase, as well as single- or double-O-deacylation followed by

phosphodiesterase activity on the resulting lysophosphatidyl-

or glycero-phospho-N-acyl ethanolamines (GP-NAEs).19

Direct release from NAPEs was found to be activated by

millimolar concentrations of calcium in vitro, and this D-type

phospholipase activity could be detected in multiple rodent

tissues, from where it was ultimately purified and molecularly

characterized.20–22 This enzyme, dubbed NAPE-PLD, was

found to be exquisitely selective for NAPEs in vitro and highly

expressed in tissues that utilize endocannabinoid signaling,

such as nervous tissue and testis.

Consistent with a role for NAPE-PLD in NAE biosynthesis,

mice lacking this enzyme [NAPE-PLD(�/�) mice] display

profound reductions in very long-chain saturated and mono-

unsaturated NAEs in the central nervous system.23 Interestingly,

however, poly-unsaturated NAEs, including AEA, were

unaltered in the nervous system of NAPE-PLD(�/�) mice.

This finding engendered the hypothesis that multiple enzy-

matic pathways are involved in the conversion of NAPEs to

NAEs in vivo, with NAPE-PLD controlling the biosynthesis of

long-chain saturated NAEs, and additional enzymes partici-

pating in the generation of shorter chain and poly-unsaturated

NAEs (including AEA). Attempts to detect an additional

D-type phospholipase activity in NAPE-PLD(�/�) tissues

were unsuccessful, but robust O-deacylation of NAPE was

observed, suggesting that lysoNAPEs or GP-NAEs might

serve as direct precursors to NAEs.24 In support of this

deacylation hypothesis, it was observed that inhibitors of

A/B-type phospholipases blocked the release of NAE

from NAPE and lysoNAPE in brain homogenates from

NAPE-PLD(�/�) mice indicating that total O-deacylation

must occur prior to phosphodiester bond cleavage under these

in vitro conditions.

Recent work from our lab has identified candidate enzymes

that can perform each step in this proposed deacylation–

phosphodiesterase pathway for NAE biosynthesis. The previously

uncharacterized a-b hydrolase enzyme ABH4 acts as a B-type

phospholipase selective for (lyso)NAPE and produces GP-NAE

from NAPE in vitro. The resulting GP-NAE is a substrate for

the metal-dependent enzyme GDE125 (glycerophosphodiester

phosphodiesterase 1, also known as MIR16), which was

previously characterized in vitro as a glycerophosphodiesterase

that acts on multiple glycerophosphodiester substrates.26,27 To

test the contribution of this pathway to NAE biosynthesis

in vivo, we herein report the generation and characterization of

mice bearing a targeted disruption of the GDE1 gene.

Results and discussion

Generation of GDE1(�/�) mice

Construction of mice bearing a disruption in the GDE1 gene

was achieved using a standard targeting strategy directly in

embryonic stem (ES) cells from the C57BL/6 background. A

targeting construct was designed to disrupt exon two of the

GDE1 gene locus, which contains residues that are required

for catalytic activity (Fig. 1A). A homologously recombined

ES-cell clone was identified by Southern blotting (Fig. 1B)

and used to generate chimeric mice on an albino C57BL/6

background. These chimeric mice gave germline transfer of

the mutated locus (Fig. 1C), and the resulting pups served

as founders for the GDE1(�/�) strain (Fig. 1D). Western

blotting of tissue extracts from GDE1(�/�) brains confirmed

loss of the GDE1 protein (Fig. 1E).

Characterization of GP-NAE and NAE metabolism in

GDE1(�/�) mice

GDE1(�/�) mice were born at the expected Mendelian

frequency, were viable and healthy, and showed no overt

differences in their cage behavior compared to wild-type litter-

mates. Initial biochemical characterization of the GDE1(�/�)mice focused on measurement of GP-NAE phosphodiesterase

activity in brain homogenates (Fig. 2A). These experiments

revealed that GDE1 is responsible for nearly all (>95%) of

the detectable GP-NAE phosphodiesterase activity in brain

tissue. Additionally, we found that conversion of lysoNAPE to

NAE was also impaired in tissue extracts (Fig. 2B), confirming

that conversion of lysoNAPE to NAE proceeds through a

GP-NAE intermediate in vitro. Previously, we described an experi-

ment in which brain homogenates accumulated ‘‘endogenous’’

GP-NAEs in an MAFP-sensitive, EDTA-induced manner,

and we hypothesized that GDE1, a metal-dependent enzyme,

was the EDTA-sensitive phosphodiesterase responsible for

this accumulation.25 To test this hypothesis, a similar experi-

ment was performed in which brains from GDE1(+/�) and(�/�) mice were incubated with or without EDTA and the

accumulation of GP-NAE was measured by LC-MS/MS

(Fig. 2C). Consistent with previous studies, GP-NAEs accumu-

lated in brain tissue from GDE1(+/�) mice in an EDTA-

dependent manner. Interestingly, GP-NAE levels in GDE1(�/�)brains were basally high after 4 hour incubation and equivalent

in magnitude to those observed in EDTA-treated GDE1(+/�)brains. No additional EDTA-induced accumulation of

GP-NAEs was observed in GDE1(�/�) brains. These results

confirm that GDE1 is the principal EDTA-sensitive GP-NAE

phosphodiesterase in brain homogenates.

To determine if the deacylation–phosphodiesterase pathway

is operant in vivo, we undertook targeted measurements of

endogenous metabolites via LC-MS/MS. First, we measured

GP-NAEs in the brains of GDE1(+/�) and (�/�) animals but,

surprisingly, observed no significant differences between these

genotypes (Fig. 3). Basal NAE levels were also unchanged in

GDE1(�/�) brains (Fig. 4).

Characterization of GP-NAE and NAE metabolism

in GDE1(�/�), NAPE-PLD(�/�), and GDE1(�/�)/NAPE-PLD(�/�) mice

As multiple enzymes are known to utilize NAPE as a substrate

(Fig. 5A), we next crossed GDE1 and NAPE-PLD lines to

generate GDE1(�/�)/NAPE-PLD(�/�) double knockout

mice to measure the combined effects of ablation of these

enzymes on NAE metabolism. Brain tissue from wild-type,

double-knockout, and each single-knockout was assayed for

the ability to release 14C-NAE from 14C-NAPE. As has been

previously reported,23 the choice of assay conditions affected

1412 | Mol. BioSyst., 2010, 6, 1411–1418 This journal is �c The Royal Society of Chemistry 2010

the relative contributions of these two pathways, with the

presence or absence of calcium proving particularly dramatic.

Furthermore, we found that selection of reaction buffer and

inclusion of detergents also influenced the relative contribu-

tions made by each pathway (Fig. S1 and S2, ESIw). Depending

on the assay conditions employed, absence of either NAPE-PLD

or GDE1 resulted in major reductions in NAPE-to-NAE

conversion, but, under all conditions tested, the double-knockout

showed near-total loss of NAPE-to-NAE conversion in brain

homogenates (Fig. 5B). Similar results were obtained in testis

homogenates (not shown).

Despite the near-complete loss of NAPE-to-NAE conver-

sion in brain homogenates from GDE1(�/�)/NAPE-PLD(�/�)double knockout mice, no significant differences in bulk brain

levels of NAEs were observed between NAPE-PLD(�/�) andthe GDE1(�/�)/NAPE-PLD(�/�) mice (Fig. 4). We next

asked whether the rate of NAE production might be impaired

by GDE1 and/or NAPE-PLD disruption by treating mice with

PF-3845, a recently described inhibitor of the NAE-degrading

enzyme fatty acid amide hydrolase (FAAH).28 FAAH inhibition

has previously been shown to produce a steady and dramatic

rise in NAE levels that plateau after about 2–3 hours.29,30 We

therefore measured brain NAE levels in various mouse models

at 3 hours following treatment with PF-3845. Interestingly, at

this time point, GDE1(�/�)/NAPE-PLD(�/�) mice displayed

significant reductions in several NAEs including AEA (Fig. 6).

The magnitude of these reductions was rather modest

(B25–40%), but clearly suggests that GDE1 and NAPE-PLD,

together, contribute to NAE biosynthesis. Similar effects on

NAE accumulation were observed in testis from these mice

(Fig. S3, ESIw). Furthermore, the fact that these reductions

were only observed following disruption of both GDE1 and

NAPE-PLD indicates that these pathways have the potential

to crosstalk with one another to regulate NAE production.

Evidence for additional NAPE-to-NAE enzymatic pathways in

neurons

Taken together, our results with GDE1(�/�) and NAPE-

PLD(�/�) mice are somewhat perplexing. Mice lacking both

GDE1 and NAPE-PLD display wild-type levels of NAEs and

only moderately compromised rates of NAE production,

Fig. 1 Generation of GDE1(�/�) mice. (A) The genomic structure of the GDE1 gene locus on chromosome 7 is shown, along with the targeting

construct used to delete exon 2 (which contains several residues essential for catalysis, Glu97, Asp99 and His112; asterisk) and the final recombined

locus. Only relevant restriction sites are designated. The locations of DNA probes used for Southern blotting 50 and 30 to the targeting con-

struct are also shown. (B) Southern blot of EcoRV-digested ES cell DNA showing the heterozygous (targeted) clone #301. (C) Southern blot of

BspHI/XhoI-digested tail-DNA from heterozygous and wild-type mice demonstrating germ-line transmission. (D) PCR strategy in which the

wild-type locus is identified by a 238 bp band and the targeted locus is identified by a 417 bp band. (E) Western blot of the particulate fraction from

brain tissue confirming loss of GDE1 protein. GDE1 is a PNGaseF-sensitive glycosylated protein.

This journal is �c The Royal Society of Chemistry 2010 Mol. BioSyst., 2010, 6, 1411–1418 | 1413

despite a total loss of NAE release from NAPE in brain

homogenates. Two possible explanations for this situation

present themselves: (1) NAEs do not, in fact, come from

NAPEs in vivo, or (2) NAEs are released from NAPEs

in vivo by an enzyme or enzymes whose activities are not

maintained in vitro. To explore the latter possibility, we sought

to develop a live-cell system in which the conversion of NAPE

to NAE could be monitored in situ. To achieve this goal, we

cultured primary neurons from neonatal wild-type, GDE1(�/�),NAPE-PLD(�/�), or GDE1(�/�)/NAPE-PLD(�/�) pups

and grew them for 6 days in vitro. Neurons were then treated

with the FAAH inhibitor PF-3845 (to prevent turnover of

NAE) and 14C-NAPE was added to the media. In this way, we

observed clear time-dependent release of NAE from NAPE in

live neurons (Fig. 7A). Previous studies have monitored the

formation of NAE in neurons by employing either radio-

labeled ethanolamine or arachidonate,7,31 which do not provide

information on the direct precursor of NAE biosynthesis.

Under our experimental conditions, the presence of a FAAH

inhibitor prevented hydrolysis of the amide bond linking14C-fatty acid to phosphatidylethanolamine thereby ensuring

that all NAE formed must be released from NAPE. Once this

experimental paradigm was established, we tested the ability of

neurons from single- and double knockout animals to release

NAE from NAPE. To our surprise, we found that all geno-

types including GDE1(�/�)/NAPE-PLD(�/�) released NAE

from NAPE at equivalent levels (Fig. 7B and C). This result

implies the existence of an additional enzyme(s) distinct from

NAPE-PLD and GDE1 that can release NAEs from NAPEs.

To test whether this in situ-active enzymatic activity requires

O-deacylation or whether it operates directly on NAPE, we

employed a non-hydrolyzable analogue of NAPE that

contains ether-linkages in the sn-1 and sn-2 positions.24 This

ether–NAPE analogue did not serve as an effective substrate

for NAE formation (Fig. 7D), suggesting that the in situ

(GDE1- and NAPE-PLD-independent) pathway for NAE

generation requires O-deacylation prior to phosphodiester

bond cleavage.

Conclusion

In summary, brain tissue from mice lacking GDE1 and

NAPE-PLD shows a near-complete loss in NAPE conversion

to NAE. Despite this result, however, bulk brain levels of

NAEs were unaltered in mice lacking GDE1 and NAPE-PLD.

A defect in the rate of NAE production was observed in

Fig. 2 GP-NAE phosphodiesterase activity is lost in GDE1(�/�) tissue. (A) Near-complete loss of conversion of 14C-GP-NAE to 14C-NAE is

evident from a representative thin layer radiochromatogram (left) and quantified data from replicate experiments (right). (B) Loss of conversion of14C-lysoNAPE to 14C-NAE is evident from a representative thin layer radiochromatogram (left) and quantified data from replicate experiments

(right). Note also the accumulation of the GP-NAE intermediate in the assay with GDE1(�/�) tissue. All assays were performed in PBS buffer

with tissue homogenates from brains of GDE1(+/+) and (�/�) mice. (C) Ex vivo assay demonstrating that EDTA induces GP-NAE

accumulation in brain homogenates due to disruption of GDE1 activity. Brain tissue from GDE1(+/�) and (�/�) mice was homogenized in

the presence or absence of 10 mM EDTA (which inhibits GDE1) and incubated at room temperature for 4 hours. The addition of EDTA leads

to a rapid accumulation of GP-NAEs in wild-type brain, as previously reported. GP-NAE accumulation was observed in brain tissue from

GDE1(�/�) mice with or without EDTA, demonstrating that GDE1 is the EDTA-sensitive GP-NAE phosphodiesterase in wild-type tissue. n= 4

mice per group. Results are presented as means� standard error. Asterisks designate po 0.05, po 0.01, po 0.005 (Student’s t-test) for *, **, and

***, respectively. Asterisks represent difference from wild-type in panels (A) and (B), and difference from EDTA-treated GDE1(+/�) tissue in

panel (C).

Fig. 3 GP-NAE levels are unchanged in GDE1(�/�) brains. Endo-genous GP-NAEs were measured in brain extracts from GDE1(+/�)and (�/�) mice via LC-MS/MS. No significant differences were

observed. n = 8 mice per group. Results are presented as means �standard error.

1414 | Mol. BioSyst., 2010, 6, 1411–1418 This journal is �c The Royal Society of Chemistry 2010

GDE1(�/�)/NAPE-PLD(�/�) mice, suggesting that both

GDE1 and NAPE-PLD make partial contributions to the

biosynthesis of anandamide and other NAEs in vivo. Con-

sidering further that all of our measurements were made

with bulk brain tissue, it is also possible that GDE1 and/or

NAPE-PLD play even more substantial roles in AEA/NAE

biosynthesis in specific neuronal circuits or brain regions.

When NAPE metabolism was studied in live neurons

(‘‘in situ’’), we discovered that neurons lacking both GDE1

and NAPE-PLD are capable of converting NAPEs to NAEs.

This result supports the existence of additional NAPE-

catabolizing enzymes that are operant in situ, and likely in vivo,

but artificially inactivated in brain homogenates. When con-

sidering the history of other phosphodiesterase enzymes, one

finds that this situation is not exceptional. The characteriza-

tion of the prototypical mammalian phospholipase D (PLD1)

involved years of frustration during which activity in intact cell

systems was observed, but lost upon cell-lysis. In fact, several

reports similar to this one can be found in the literature in

which PLD activity was measured in intact cell prepara-

tions, but could not be reconstituted in vitro.24,32,33 The

breakthrough discovery came when it was found that PLD1

activity was dramatically enhanced in vitro by performing

substrate assays under precise conditions: addition of the

non-hydrolyzable GTP analogue GTPgS, addition of the

ADP-ribosylation factor (ARF) protein, as well as substrate

presentation in mixed liposomes that include phospha-

tidylinositol-4,5-bisphosphate (PIP2).34 Indeed, omission of

Fig. 4 Bulk NAE levels in brains from mice lacking GDE1 and/or NAPE-PLD. Brains from wild-type, GDE1(�/�), NAPE-PLD(�/�) orGDE1(�/�)/NAPE-PLD(�/�) mice were subjected to organic extraction and endogenous NAE levels were measured by LC-MS/MS. No

significant differences were observed between wild-type and GDE1(�/�) or between NAPE-PLD(�/�) and GDE1(�/�)/NAPE-PLD(�/�)brains.

Fig. 5 NAPE conversion to NAE in brain lysates from mice lacking GDE1 and/or NAPE-PLD. (A) Pathway diagram showing potential routes

for NAPE conversion to NAE. Note that both NAPE-PLD and GDE1 can liberate NAE from its esterified precursors (NAPE and GP-NAE,

respectively). (B) Brain tissue from wild-type, GDE1(�/�), NAPE-PLD(�/�), or GDE1(�/�)/NAPE-PLD(�/�) mice was homogenized and

assayed for the ability to convert NAPE to NAE in PBS buffer in the presence or absence of 10 mM CaCl2 or EDTA via thin layer

radiochromatography. In the presence of calcium or EDTA, tissue lacking NAPE-PLD shows complete loss of NAPE-to-NAE conver-

sion. However, assays performed without these additives show no impairment from loss of NAPE-PLD and a large reduction in activity in

tissue lacking GDE1. Under all conditions, tissue from the GDE1(�/�)/NAPE-PLD(�/�) displayed negligible activity. The results are presented

as means � standard error. n = 4 mice per group. The symbols above the bars designate p o 0.05 (Student’s t-test) with * for different from

wild-type, # for different from GDE1(�/�) samples and w for different from NAPE-PLD(�/�) samples.

This journal is �c The Royal Society of Chemistry 2010 Mol. BioSyst., 2010, 6, 1411–1418 | 1415

any of these factors resulted in almost complete loss of

activity. That enzyme, PLD1, is selective for phosphatidyl-

choline (PC), although to our knowledge it has not been tested

with N-acylated substrates. Additionally, there are another six

PLD isoforms in mammalian genomes, five of which are

virtually unstudied due to inability to show activity in vitro.

Furthermore, there are several GDE1 homologues that, to our

knowledge, have not yet been shown to display activity

in vitro. Finally, other pathways have been proposed for

NAPE catabolism involving C-type phospholipase (PLC)

activity (resulting in phospho-NAE, which is a substrate for

one or more phosphatases),35,36 although direct measurement

of NAPE-PLC activity has not yet been described. This does

not necessarily argue against a PLC pathway, since mammalian

C-type phospholipases have also been shown to have complex

requirements for co-factors and protein activators for reconsti-

tution of activity in vitro.37 Looking forward, the ability of

various co-factors to stimulate endogenous NAPE catabolic

pathway(s) should be explored to more clearly define their

composition (e.g., do they involve A/B-, C-, or D-type phospho-

lipases, or some combination thereof?) and possible role in

AEA biosynthesis in vivo. Finally, considering the recent

discovery of several classes of endogenous N-acyl amino acids

in brain tissue,38 it is worth investigating whether alternative

Fig. 6 Impaired NAE accumulation is observed following FAAH inhibition in brains from mice lacking GDE1 and NAPE-PLD. Mice were

treated with the selective FAAH inhibitor PF-3845 (10 mg kg�1, i.p.) and, after 3 hours, sacrificed and brain NAE levels measured. Significant

reductions were observed for several NAEs, including the endocannabinoid AEA (C20:4), in brains from GDE1(�/�)/NAPE-PLD(�/�) mice.

The results are presented as means � standard error. n = 4 mice per group. The symbols above the bars designate p o 0.05, p o 0.01, p o 0.005

(Student’s t-test) with *, **, *** for different from wild-type, #, ##, ### for different from GDE1(�/�) samples and w, ww, www for different from

NAPE-PLD(�/�) samples, respectively.

Fig. 7 In situNAE release fromNAPE in primary neuron cultures frommice lacking GDE1 andNAPE-PLD. (A) Time-dependence of NAE release

in the presence or absence of the FAAH inhibitor PF-3845. NAE release in the absence of neurons was negligible on this timescale (not shown).

(B) Assays conducted in primary neurons for 6 hours in the presence of PF-3845. Comparison of wild-type neurons to neurons from GDE1(�/�) orNAPE-PLD(�/�) mice reveals no impairment in the ability to release NAE from NAPE in situ. (C) Comparison of NAPE-PLD(�/�) neurons toGDE1(�/�)/NAPE-PLD(�/�) neurons reveals wild-type levels of NAE release in GDE1(�/�)/NAPE-PLD(�/�) neurons. (D) NAE release

from 14C-NAPE or an ether-linked (non-hydrolyzable) NAPE analogue in neurons from GDE1(�/�)/NAPE-PLD(�/�) mice reveals the

necessity of de-acylation prior to NAE release. These data represent single experiments performed in triplicate, representative of multiple experiments.

*, p o 0.05.

1416 | Mol. BioSyst., 2010, 6, 1411–1418 This journal is �c The Royal Society of Chemistry 2010

pathways for AEA/NAE production may exist that do not

involve NAPEs as obligate precursors.

Experimental

Generation of GDE1(�/�) mice

All animal experiments were performed in compliance with

protocols approved by The Scripps Research Institute’s Institu-

tional Animal Care and Use Committee. The GDE1 locus was

obtained as part of a BAC clone from the RPCI-23 library of

genomic DNA from a female C57BL/6J mouse. The targeting

construct was generated using PCR-amplified 50 and 30 homo-

logous recombination fragments surrounding exon 2 of theGDE1

gene, which were subcloned into NotI/EcoRI and XhoI/HindIII

sites in the pKO-NTKV vector. Homologous recombinant Bruce-

4 stem cell clones (C57BL/6 strain of origin) were identified by

Southern blot analysis, and one such clone was used to generate

chimeric mice on an albino C57BL/6 background, several of

which yielded germline transmission of the mutated gene.

In vitro enzyme assays

All assays were performed in tissue homogenates. Mice were

anesthetized with isoflurane and killed by decapitation. Tissues

were immediately removed and snap-frozen in liquid N2. Frozen

tissue was homogenized in phosphate buffered saline (PBS,

except where noted) using a Dounce-homogenizer and centri-

fuged at 1000� g to remove tissue debris. The supernatant from

this centrifugation step was adjusted to 2 mg mL�1 with PBS

and stored in single-use aliquots at �80 1C. Except where noted,assays were performed in glass vials in PBS at 37 1C for 30 min

in 100 mL total volume. Final protein concentration for enzyme

assays was 1 mg mL�1 for NAPE and lysoNAPE substrates,

and 0.1 mg mL�1 for GP-NAE substrates. Substrates (1,2-

dioleoyl-sn-glycero-3-phospho(N-[10-14C]-palmitoyl)ethanolamine,

1-oleoyl-2-hydroxy-sn-glycero-3-phospho(N-[10-14C]-palmitoyl)-

ethanolamine, and 1,2-dihydroxy-sn-glycero-3-phospho(N-[10-14C]-

palmitoyl)ethanolamine) were prepared as previously described.24,25

Substrates were stored as 2 mM stock in EtOH at �80 1C

and 5 mL was added to each enzyme assay (for 100 mM final

concentration). After 30 min, 1.5 mL of 2 : 1 CHCl3 : MeOH

was added to quench the reactions and an additional 0.4 mL

of 1% formic acid was added. Vials were mixed and

briefly centrifuged to separate phases and the organic

(bottom) phase was removed to fresh vials and evaporated

to dryness under N2. Lipids thus extracted were resuspended

in 20 mL 2 : 1 CHCl3 : MeOH and spotted on thin layer silica

plates and developed in 40 : 10 : 1 CHCl3 : MeOH : 28%

NH4OH. Distribution of radioactivity on the plates was quantified

by a phosphorimaging device (Packard) and products were identi-

fied by comparison to synthetic standards. Heat-denatured or

no-protein controls were included with all assays and these values

were subtracted as background. As noted in the text, some assays

included one of the following additives: 10 mM EDTA, 10 mM

CaCl2 or 0.1% Triton X-100.

Primary neuron culture and in situ assays

Culturing of primary neurons was performed essentially as

previously described.39 Briefly, P0 pups were taken from their

mothers and kept at B37 1C prior to dissection. Pups were

tailed and PCR-genotyped within 2 h of separation from their

mothers (Fig. 1D). Pups with the desired genotypes were

sacrificed and forebrains were dissected and dissociated by

incubation for 7 min at 37 1C in digestion solution containing

6 mg mL�1 trypsin, 0.5 mg mL�1 DNAse, 137 mM NaCl,

5 mM KCl, 7 mM Na2HPO4, and 25 mM HEPES, pH 7.2.

The dissociated cells were washed once with Hank’s balanced

salt solution (HBS) containing 20% fetal bovine serum (FBS)

and once in serum-free HBS, and further dissociated by gentle

trituration in HBS containing 12 mM MgSO4 and 0.5 mg mL�1

DNAse. The cell suspension was centrifuged for 3 min

at 1000 � g and plated in Matrigel-coated 6-well dishes

in MEM supplemented with glutamine, insulin, transferrin,

glucose, and 10% FBS. Cells from the forebrain of a single

pup were used to plate 6 wells in a 6-well dish. After B12 h,

the media was removed and replaced with MEM supple-

mented with glucose, transferrin, B27, glutamine, and 5%

FBS (‘‘growth media’’). After 3 days of culture (when glial

cells reached B40–50%), 50% of the culture-medium was

replaced with fresh growth medium containing 10 mM Ara-C

(for 5 mM final concentration) to prevent glial proliferation.

The cultures were maintained in medium containing 5 mMAra-C. After 6 days in vitro, 2 mL PF-3845 in DMSO or

DMSO was added to a final concentration of 10 mM and cells

were returned to the incubator. After 1 h, media was removed,

and replaced with 1 mL of fresh growth media containing

10 mMPF-3845 as well as 14C-labeled substrate at 20 mM (final

concentration), and neurons were returned to the incubator

for various time points, typically 6 h. After 6 h, each well was

scraped and neurons and media were transferred to a glass vial

containing 6 mL of 2 : 1 CHCl3 : MeOH and 1 mL of 1%

formic acid for 8 mL total. Vials were vortexed to mix,

centrifuged to separate phases, and the organic (bottom) phase

was removed to a fresh glass vial and evaporated to dryness

under a stream of N2. Dried reactions were then analyzed by

thin layer radiochromatography as described above.

Brain-lipid extraction and tandem MS analysis of NAE and

GP-NAE

Endogenous lipid measurements were performed essentially as

previously described. Briefly, male mice between 2 and 6

months of age were anesthetized with isoflurane and killed

by decapitation. Brains were rapidly removed, sectioned into

left and right hemispheres, and immediately frozen in liquid

N2. Frozen brains were placed in a Dounce homogenizer with

8 mL of 2 : 1 : 1 CHCl3 : MeOH : 50 mM Tris, pH 8.0, and

200 pmol of 1,2-dihydroxy-sn-glycero-3-phospho(N-penta-

decenoyl)ethanolamine (C15:1 GP-NAE) or 2 pmol of arachi-

donoyl-(d4)-ethanolamine for GP-NAE or NAEmeasurements,

respectively. Lipids were extracted via homogenization with

8–10 strokes of the Dounce homogenizer at which point

homogenates were transferred to 8 mL glass vials and centri-

fuged for 10 min at 1400 � g. The organic (bottom) phase was

transferred to a clean vial and the remaining aqueous

phase was re-extracted with an additional 4 mL of CHCl3.

After centrifugation, the organic phases were pooled and

evaporated to dryness under a stream of N2. Lipid extracts

This journal is �c The Royal Society of Chemistry 2010 Mol. BioSyst., 2010, 6, 1411–1418 | 1417

from each half-brain were then resuspended in 120 mL 2 : 1

CHCl3 : MeOH and transferred to LC-MS vials for immediate

analysis or stored at �80 1C. 15 mL of resuspended lipid was

injected into an Agilent 6410 triple quadrupole mass spectro-

meter with an Agilent 1200 autosampler. Samples were

chromatographed on a Gemini 5 mm C18 column (50 � 4.6 mm,

Phenomenex, Torrance, CA) in the negative-ion mode for

GP-NAE measurements and the positive-ion mode for NAE

measurements at a flow-rate of 0.4 mL min�1. GP-NAEs and

NAEs were monitored by multiple reaction monitoring

(MRM). For GP-NAEs the transition from [M � H]� to

[M � H � 74]� with collision energy 35 V was monitored

and compared to the internal C15:1 GP-NAE standard and

measurements of endogenous GP-NAEs were corrected for

extraction efficiency as previously described.25 For NAEs, the

MRM transition from [M + H]+ to m/z 62+ (ethanolamine

fragment) with collision energy 11 V was used, except for the

d4-AEA standard which monitored the transition from

[M + H]+ to m/z 66+ (d4 ethanolamine fragment). NAEs

were quantified by comparison to the d4-AEA standard. The

dwell-time was set to 100 ms for all lipids. For ex vivo analysis

of GP-NAEs, half-brains were homogenized in a Dounce

homogenizer in 2 mL of 50 mM Tris pH 8.0 with or without

10 mM EDTA. Samples were incubated at room temperature

for 4 h with gentle rotation prior to extraction and GP-NAE

quantification as described above.

References

1 P. Pacher, S. Batkai and G. Kunos, Pharmacol. Rev., 2006, 58,389–462.

2 A. C. Howlett, C. S. Breivogel, S. R. Childers, S. A. Deadwyler,R. E. Hampson and L. J. Porrino, Neuropharmacology, 2004, 47,345–358.

3 L. Walter and N. Stella, Br. J. Pharmacol., 2004, 141, 775–785.4 C. J. Fowler, S. Holt, O. Nilsson, K. O. Jonsson, G. Tiger andS. O. Jacobsson, Pharmacol., Biochem. Behav., 2005, 81, 248–262.

5 K. Ahn, M. K. McKinney and B. F. Cravatt, Chem. Rev., 2008,108, 1687–1707.

6 V. Di Marzo, Rev. Physiol., Biochem., Pharmacol., 2008, 160, 1–24.7 H. Cadas, S. Gaillet, M. Beltramo, L. Venance and D. Piomelli,J. Neurosci., 1996, 16, 3934–3942.

8 C. J. Fowler, Fundam. Clin. Pharmacol., 2006, 20, 549–562.9 S. T. Glaser, M. Kaczocha and D. G. Deutsch, Life Sci., 2005, 77,1584–1604.

10 D. G. Deutsch, N. Ueda and S. Yamamoto, Prostaglandins,Leukotrienes Essent. Fatty Acids, 2002, 66, 201–210.

11 H. H. Schmid, Chem. Phys. Lipids, 2000, 108, 71–87.12 W. A. Devane, L. Hanus, A. Breuer, R. G. Pertwee,

L. A. Stevenson, G. Griffin, D. Gibson, A. Mandelbaum,A. Etinger and R. Mechoulam, Science, 1992, 258, 1946–1949.

13 C. C. Felder, E. M. Briley, J. Axelrod, J. T. Simpson, K. Mackieand W. A. Devane, Proc. Natl. Acad. Sci. U. S. A., 1993, 90,7656–7660.

14 D. E. Epps, P. C. Schmid, V. Natarajan and H. H. O. Schmid,Biochem. Biophys. Res. Commun., 1979, 90, 628–633.

15 V. Natarajan, P. C. Schmid, P. V. Reddy, M. L. Zuzarte-Augustinand H. H. O. Schmid, J. Neurochem., 1983, 41, 1303–1312.

16 M. Colodzin, N. R. Bachur, H. Weissbach and S. Udenfriend,Biochem. Biophys. Res. Commun., 1963, 10, 165–170.

17 D. G. Deutsch and S. A. Chin, Biochem. Pharmacol., 1993, 46,791–796.

18 T. Sugiura, S. Kondo, A. Sukagawa, T. Tonegawa, S. Nakane,A. Yamshita and K. Waku, Eur. J. Biochem., 1996, 240,53–62.

19 V. Natarajan, P. C. Schmid, P. V. Reddy and H. H. Schmid,J. Neurochem., 1984, 42, 1613–1619.

20 Y. Okamoto, J. Morishita, K. Tsuboi, T. Tonai and N. Ueda,J. Biol. Chem., 2003, 279, 5298–5305.

21 N. Ueda, Q. Liu and K. Yamanaka, Biochim. Biophys. Acta, Mol.Cell Biol. Lipids, 2001, 1532, 121–127.

22 J. Wang, Y. Okamoto, J. Morishita, K. Tsuboi, A. Miyatake andN. Ueda, J. Biol. Chem., 2006, 281, 12325–12335.

23 D. Leung, A. Saghatelian, G. M. Simon and B. F. Cravatt,Biochemistry, 2006, 45, 4720–4726.

24 G. M. Simon and B. F. Cravatt, J. Biol. Chem., 2006, 281,26465–26472.

25 G. M. Simon and B. F. Cravatt, J. Biol. Chem., 2008, 283,9341–9349.

26 B. Zheng, D. Chen and M. G. Farquhar, Proc. Natl. Acad. Sci.U. S. A., 2000, 97, 3999–4004.

27 B. Zheng, C. P. Berrie, D. Corda and M. G. Farquhar, Proc. Natl.Acad. Sci. U. S. A., 2003, 100, 1745–1750.

28 D. S. Johnson, K. Ahn, S. Kesten, S. E. Lazerwith, Y. Song,M. Morris, L. Fay, T. Gregory, C. Stiff, J. B. Dunbar, Jr.,M. Liimatta, D. Beidler, S. Smith, T. K. Nomanbhoy andB. F. Cravatt, Bioorg. Med. Chem. Lett., 2009, 19, 2865–2869.

29 K. Ahn, D. S. Johnson, M. Mileni, D. Beidler, J. Z. Long,M. K. McKinney, E. Weerapana, N. Sadagopan, M. Liimatta,S. E. Smith, S. Lazerwith, C. Stiff, S. Kamtekar, K. Bhattacharya,Y. Zhang, S. Swaney, K. Van Becelaere, R. C. Stevens andB. F. Cravatt, Chem. Biol., 2009, 16, 411–420.

30 D. Fegley, S. Gaetani, A. Duranti, A. Tontini, M. Mor,G. Tarzia and D. Piomelli, J. Pharmacol. Exp. Ther., 2004, 313,352–358.

31 V. Di Marzo, A. Fontana, H. Cadas, S. Schinelli, G. Cimino,J.-C. Schwartz and D. Piomelli, Nature, 1994, 372, 686–691.

32 S. Cockcroft, Biochim. Biophys. Acta, 1992, 1113, 135–160.33 T. W. Martin and K. Michaelis, J. Biol. Chem., 1989, 264,

8847–8856.34 H. A. Brown, S. Gutowski, C. R. Moomaw, C. Slaughter and

P. C. Sternwels, Cell (Cambridge, Mass.), 1993, 75, 1137–1144.35 J. Liu, L. Wang, J. Harvey-White, B. X. Huang, H.-Y. Kim,

S. Luquet, R. D. Palmiter, G. Krystal, R. Rai, A. Mahadevan,R. K. Razdan and G. Kunos, Neuropharmacology, 2008, 54,1–7.

36 J. Liu, L. Wang, J. Harvey-White, D. Osei-Hyiaman, R. Razdan,Q. Gong, A. C. Chan, Z. Zhou, B. X. Huang, H. Y. Kim andG. Kunos, Proc. Natl. Acad. Sci. U. S. A., 2006, 103, 13345–13350.

37 A. V. Smrcka, J. R. Hepler, K. O. Brown and P. C. Sternweis,Science, 1991, 251, 804–807.

38 B. Tan, D. K. O’Dell, Y. W. Yu, M. F. Monn, H. V. Hughes,S. Burstein and J. M. Walker, J. Lipid Res., 2009,M900198–JLR900200.

39 A. Maximov, Z. P. Pang, D. G. R. Tervo and T. C. Sudhof,J. Neurosci. Methods, 2007, 161, 75–87.

1418 | Mol. BioSyst., 2010, 6, 1411–1418 This journal is �c The Royal Society of Chemistry 2010